Narrative-scope diligence
AI Investor Review: How Large Language Models Evaluate Your Narrative
The story you present and the story models hear when placed in an evaluator role are not always the same.
A pitch deck, business plan, or sale memorandum is written for a deciding audience. Large language models can be conditioned on that same audience role — investment reviewer, credit officer, acquirer analyst — and asked to evaluate the narrative against a fixed rubric rather than merely describe the company.
The output is a structured preview of reception: overall narrative score, rubric-level strengths and weaknesses, and the disagreements between models about the strength of the story. In operational experience the reservations that surface are frequently the same reservations human reviewers later raise. For companies preparing materials the review is advance notice. For investors, credit teams, and acquirers it is a way to see how the AI layer their own analysts already consult receives the same document.
Live report surface
AI Investor Review · narrative-scope
Northline Analytics · illustrative fixture on the real UI. Not a screenshot.
AI Investor Review
Investor Lens rubric — 5 parameters
72/100
Investor Readiness
Top 3 fixes
- Add current ARR, paid customers, or signed pilots on a dated traction slide.
- State the raise amount, instrument, and a 100% use-of-funds split.
- Replace vanity user counts with an active / paid breakdown.
What the module actually does
Evaluator role, fixed rubric, multi-model scoring
The narrative-review module applies the same multi-model apparatus used for brand assessment, with two decisive differences.
First, the models are conditioned on an explicit evaluator role matched to the document’s intended audience. An investor-facing deck is reviewed as an investment reviewer would review it; a business plan submitted for credit is reviewed in a credit-officer frame; materials prepared for a sale process are reviewed as an acquirer’s analyst would review them.
Second, the models score the narrative against a fixed rubric rather than answering open descriptive queries. The rubric scores are aggregated into an overall narrative score that is reported separately from the brand-scope VeritasScore.
Because the story is scored by its own rubric, brand-scope and combined assessments of the same firm on the same date can differ. The difference locates in the evaluative layers — the synthetic panel and the narrative review — rather than in simple mention or sentiment signals. That separation is intentional: it distinguishes “how models describe the company” from “how models evaluate the account the company gives of itself.”
Methodologically the module inherits the same disciplines as brand assessment: battery sizing above the noise floor and policy-bounded conversational elicitation. Outputs are therefore comparable across models and across time on the same terms.
Dual read of the output
Advance notice for companies, structured summary for evaluators
For the company preparing the materials
Advance notice
The review surfaces the reservations models raise when placed in the deciding role. Those reservations are, in operational experience, predictive of the questions and concerns human reviewers subsequently bring into the meeting. Identifying weak authority claims, unclear defensibility language, or recurring objections before the meeting allows the narrative to be strengthened where it is most exposed.
For the investor, credit officer, or acquirer
Structured summary
The same measures run prospectively over a target constitute a structured summary of how the AI layer already present in many workflows perceives the target’s narrative. A credit officer receiving a business plan can pair the document with its measured digital footprint and see both the coded reservations and the disagreements among models that a single-assistant workflow never surfaces. An investment analyst can see whether models evaluate the story more or less favorably than they describe the company in open queries.
In both readings the output is evidence, not a decision. It makes an already-active influence visible and reviewable.
Relationship to brand-scope measurement
Why the two scores are kept separate
Brand-scope measurement answers: how do models currently describe and position this company across the queries real evaluators already ask?
Narrative-scope measurement answers: how do models evaluate this specific account of category, opportunity, and viability when placed in the role the document was written for?
Keeping the scores separate prevents the strength (or weakness) of the public footprint from being conflated with the strength (or weakness) of the story. A company can have solid brand visibility and still present a narrative that models find weakly supported, or the reverse. The gap between the two scores is itself information.
Live report surface
AI Investor Review · narrative-scope
Northline Analytics · illustrative fixture on the real UI. Not a screenshot.
AI Investor Review
Investor Lens rubric — 5 parameters
72/100
Investor Readiness
Top 3 fixes
- Add current ARR, paid customers, or signed pilots on a dated traction slide.
- State the raise amount, instrument, and a 100% use-of-funds split.
- Replace vanity user counts with an active / paid breakdown.
Live report surface
VeritasScore · multi-model distribution
Northline Analytics · illustrative fixture on the real UI. Not a screenshot.
Score by LLM model
620Overall · 300–870
Practical use in diligence and preparation
Where teams place the review
- Pre-meeting preparation by the company: identify and address the highest-severity reservations before they are raised live.
- Early screening by investors or acquirers: obtain a multi-model reading of the narrative alongside brand-scope perception.
- Credit assessment: pair the business plan with both brand-scope and narrative-scope outputs so that story strength and public-footprint strength are visible separately.
- Post-meeting or portfolio review: re-run after material updates to the narrative or the public footprint to detect movement.
Test the narrative the way models will when placed in the deciding role
Free analysis
Run it on any company
Public URL. Same funnel as the homepage. No card required.
Enter a public website
Multi-model · VeritasScore · ~5 minutes
See a sample dossier →FAQ
Short clarifications — positioning and proof live in the sections above.
What document types can be submitted?+
Investor-facing decks and pages, business plans prepared for credit applications, materials prepared for a sale process, and similar narrative documents. The module is agnostic to format; the defining variable is the evaluator role on which the models are conditioned.
Is this a substitute for human review of the deck or plan?+
No. It is a structured, multi-model preview of how the AI layer already present in many workflows receives the same document. Human judgment remains primary.
How does this differ from simply asking a chatbot to “review this deck”?+
The module uses a fixed multi-model battery, a consistent rubric, policy-bounded elicitation, and separation from brand-scope signals. A single ad-hoc chat produces one uncontrolled draw; the module produces a comparable, decomposable distribution.
Can confidential materials be processed?+
Handling of confidential documents follows the same controls the firm already applies to other diligence and analysis tools. Discuss specific requirements during enterprise onboarding.
Is the narrative score comparable across companies?+
Within the same methodology vintage and rubric, yes. The score is designed for comparison of narrative reception under consistent conditions.